First release of gemBS-rs, a complete rewrite of the gemBS pipeline (apart from the mapper) in Rust bringing increased stability while maintaining the high performance of gemBS: https://github.com/heathsc/gemBS-rs.git
gemBS is a high performance bioinformatic pipeline designed for highthroughput analysis of DNA methylation data from whole genome bisulfites sequencing data (WGBS). It combines GEM3, a high performance read aligner and bs_call, a high performance variant and methyation caller, into a streamlined and efficient pipeline for bisulfite sueqnce analysis.
The manuscript describing the pipeline is available here
gemBS is licensed under GPL.
Use git clone --recursive to retrieve the complete source code including the code from external projects such as bs_call and gem3-mapper.
git clone --recursive https://github.com/heathsc/gemBS.git
- Before starting the installation of gemBS, you will need to install or check the installation of several packages.
a) gcc with development libraries b) python3, pip3, matplotlib, multiprocess c) zlib, lzma, openssl, libcurl, libncurses, wget, pigz
If you are working on a clean (fairly recent) Ubuntu installation, you can install everything required with the followiwg commands:
sudo apt-get update
sudo apt-get install -y python3 build-essential git python3-pip wget pigz
sudo apt-get install -y zlib1g-dev libbz2-dev
sudo apt-get install -y libncurses5-dev liblzma-dev libssl-dev libcurl4-openssl-dev
pip3 install matplotlib multiprocess
-
Download the gemBS distribution if you haven't already done so:
git clone --recursive https://github.com/heathsc/gemBS.git -
Use python install command:
To install to the standard system location (i.e., so that all users can use gemBS):
``python3 setup.py install``
To install to the user's home directory:
``python3 setup.py install --user``
For checking your installation follow this worked example.
Documentation can be found at gemBS.
3.5.5 Fix logging bug caused by trimming change in 3.5.3
3.5.4 Fix bug in the output of strand specific cpg txt files (not
encode Bed files) where the 'C' entry was not being printed
3.5.3 Allow for read end specific trimming in bs_call
3.5.3 Enable range checks and asserts in bs_call all target; add bs_call debug target
3.5.2 Correct problems with gcc10. Move to htslib/samtools/bcftools version 1.11
3.5.1 Check if C compiler requires --std=c99 flag for standards conformant behaviour
3.5.1 Make sure bgzip is copied correctly during installation
3.5.0 Make bs_call process contig pools from largest to smallest (this change alters the sqlite db format so
if you have a previously started gemBS run you should (a) remove the .gemBS directory, (b) redo the
'gemBS prepare' step to recreate the db file and (3) run 'gemBS db-sync'.
3.5.0 Switch bs_call and snpxtr to use the new dbSNP index format
3.5.0 Add ability of dbSNP to read the new JSON and VCF dbSNP format files
that are now used for human and non-human species respectively
3.5.0 Add multithreading to dbSNP_idx
3.5.0 Change format of dbSNP index to allow (a) efficient loading
of SNP data for individual contigs and (b) parallel index creation
3.5.0 Rewrite mextr and snpxtr as standalone tools rather than
bcftools plugins. Now multithreaded and (relatively) memoryefficient
3.5.0 Replace bedToBigBed and wigToBigWig to reduce memory usage
and improve speed
3.4.5 Fix crash when using the -k (keep-mismatch) flag, and fix rare hangs at end of processing
3.4.4 Sort input bcf files to bcftools concat stage to ensure reproducibility.
3.4.4 Add extra sort keys when generating pools to ensure stability of pool membership in the event of multiple contigs
having the same size
3.4.3 Remove calculation of the goodness of filter (GOF) as this is expensive, non-standard and unreliable. Removing this
removes the dependency on GSL.
3.4.3 Add autodetection of output format to bs_call (unless explicitly specified on the command line)
3.4.2 Add CRAM support (via make_cram option in configuration file)
3.4.1 Add benchmark-mode that does not write date or program version numbers into SAM/BAM or VCF/BCF files
Switch to samtools, bcftools and htslib v1.10
3.4.0 Move to new bs_call version (2.1.0) which is more efficient
in memory use and can read BAMs and write BCFs natively.
The new bs_call requires a faidx indexed reference, so gemBS
no creates this during indexing.
3.4.0 Add switches to give more control to threads and memory
usage in mapping and calling stages
3.3.3 Remove legacy pathway for config files with no header line (fix issue 'error in gemBS index #65)
3.3.2 Fix error where header line for wig files could be omitted
3.3.2 Fix generation of non_cpg files
3.3.1 Fix Attribute error bug due to not checking if conversion is a list
3.3.0 Make new release for IHEC
3.3.0 Switch conversion default in IHEC_standard configuration to 0.01,0.05 rather than auto, which can give odd results if conversion controls not present or not working correctly
3.3.0 Fix bug where conversion parameters could be ignored
3.2.13 Fix formatting bug in mextr with multiple samples (not triggered in normal gemBS use)
3.2.12 Ensure that conversion statistics are correctly calculated for non-stranded or reverse conversion protocols
3.2.11 Introduce reverse_conversion option for mapping where read 1 is G2A converted and read 2 is C2T converted
3.2.10 Correct regex patch for single end reads
3.2.9 Update Singularity and Dockerfile recipes to allow kemp utils to be built correctly
3.2.9 Make setup.py and gemBS/commands.py read the version information from gemBS/version.py, so ensuring consistency
3.2.9 Fix bug added in last version where options in config file were not being taken into account
3.2.8 Fix mis specification errors in long options for mextr.
3.2.8 Fix bug where mextr (methyl extract plugin for bcftools) would crash if cpg output option was not set.
3.2.7 Apply patches for bugs in handling single end reads (suggested by I. Moghul)
3.2.7 Changed regex for filenames to make it more general (suggested by I. Moghul)
3.2.7 Fixed bug (reported by chhylp123) where zero arguments to some options were being ignored
3.2.6 Cleaned up compilation and cleaning of gemBS tools
3.2.6 Fixed python error if either the over conversion reference sequence was not defined
3.2.6 Added check in bs_call that conversion parameters are valid (between 0 and 1)
3.2.6 Perform more stringent sanity checking on conversion vaalues when autocomputed by gemBS
3.2.6 Use --diasble-lzma configuration flag for samtools and bcftools as we don't need it and it removes an unneccesary dependency
3.2.6 Add install options --disable-cuda (on by default) and --enable-cuda that affect GEM3 comppilation
3.2.6 Bug fix with incorrect handling of duplicate reads
3.2.5 Minor bug fix - correct error with non-paired end non-bisulfite reads
3.2.4 Modify the bisulfite processing in gem-mapper to be more efficient (in particular for the non-stranded option)
3.2.4 Modify gemBS to use the new conversion options for gem-mapper
3.2.4 Switch gem-mapper to use option --underconversion-sequence and --overconversion-sequence rather than --underconversion_sequence to be consistent with other options
3.2.3 Fixed bug if conversion parameters were not set
3.2.2 Rework non-stranded mode so that both possible conversions are tried and the results merged
3.2.2 Fix bug where non-stranded flag was not being passed to mapper in paired end mode
3.2.1 Move warning message from bscall from stdout to stderr
3.2.1 Switch Singularity build to use Ubuntu 16.04 rather than 18.04 to allow the image to work in CentOS 6 (Docker build changed as well to keep the two in sync)
3.2.1 Fix undeclared variable bugs and missing --ignore-deps option in merge-bcfs
3.2.1 Add default for dbSNP_index if dbSNP_files is set
3.2.1 Add gsl-path install option
3.2.0 Make new release
3.1.0 Make installation process more modular. Allow for sub-installs
3.1.0 Add support for reading config from ${index_dir}/gemBS.json if it exists
3.1.0 Add --reference-bias option to mextr and gemBS extract
3.1.0 Add support for non-bisulfite mapping of individual datasets
3.1.0 Allow white space in variable values
3.1.0 Allow fallback to gzip if pigz not present
3.1.0 Add --dry-run, --json, --ignore-db and --ignore-dep to extract command
3.1.0 Add --ignore-dep option to call and merge-bcfs commands
3.1.0 Add SNP extraction function to extract command
3.0 Make v3.0 release
3.0 Merge with master branch.
3.0 Bump samtools sort memory limit to 2G
3.0 Add extra_references option for reference generation
3.0 Allow input files to mapping to be shell commands
3.0 Add links to documentation
3.0 Upload new yeast example and add documentation
3.0 Add --dir option to gemBS
3.0 Add --ignore-db options for --dry-run / --json
3.0 Add --json output option for dry runs
3.0 Update help text to match new functions
3.0 Introduce standard analysis configurations stored within distribution
3.0 Switch gem3-mapper distribution to gembs branch on official gem3-mapper repo
3.0 Removal of incomplete files and roll back of db in the event of pipeline failure
3.0 Automatic removal of individual BAMs and BCFs after successful merging
3.0 Prevent pipelines hanging in event of failure
3.0 Generate ENCODE bed and bigbed files
3.0 Switch to python 3
3.0 Switch to mextr for BCF filtering
3.0 Include fetch and build of samtools / bcftools during build process
3.0 Add dry-run capability to map and call commands
3.0 Introduce contig pools to automatically group small contigs
3.0 Automatic generation of contig.size files from index build
3.0 Allow use of in memory sqlite3 db as an option
3.0 Allow multiple instances of gemBS (possible on different hosts) to work
simultaneously on the same analysis
3.0 Reduce and simply commands
3.0 Add Dockerfile
3.0 Add multi-threading and multi-processing options for most commands
3.0 Use sqlite3 to track progress of analyses, file paths etc.
3.0 Added more flexible configuration options (new csv format + new configuration file)
3.0 Remove test dataset from distribution (distribute from web site)
2.1.0 Ensure commands run during pipeline come from installation
2.1.0 Added Singularity build recipe
2.1.0 Add new command gemBS direct-mapping
2.1.0 Fixed Makefile clean in tools
2.0.2 Fixed bug related with the percentage of High Quality Variant in Variants summary report.
2.0.2 Check temporary directory existence.
2.0.2 Fixed QualityNonRefCpg sample name in png image.
2.0.2 Fixed mapper issues related with aligning performace.
2.0.2 Fixed arguments for Under/Over Conversion sequence name in gem3-mapper
2.0.1 On bscall repository, fixed argument -k about discarded reads that do not form proper pairs.
2.0 Check tmp folder before starting mapping process.
2.0 Added Left and Right Trimming optional arguments to gemBS bscall.
2.0 Added GC Coverage correlation value to BS Call Stats Summary.
2.0 Fixed error when reporting complete path to not found bam files.
2.0 Fixed iteration over sampleBams dictionary in MergeAll method.
2.0 Updated: Avoid redo indexing when merging just one file.
2.0 Changed conversion formula.
2.0 Added parameter for dbSNP.
2.0 Added threads to bscall.
2.0 Removed CpGs reports. Already done from bscall report.
2.0 Fixed bs_call makefile for the gcc to be used.
2.0 New bscall version. Generates JSON report.
2.0 Removed gemBS options snp-stats,cpg-report,cpg-stats.
2.0 Added summary report from the bs_call json stats
2.0 New BSCall Report. From bscall son file generates three types of reports:
Mapping and Coverage Report
Bs-Genotypes Calls Report
Methylation Statistics report
1.7 Added non stranded read conversion parameter
1.7 Fixed SE crash when estimating overlapped bases.
1.7 Fixed gem-index (gem3) to follow fastq and SAM specifications.
Modified gem3-mapper repository external module.
New external module https://github.com/heathsc/gem3-mapper.git
1.7 Fixed threads parameter to samtools merge
1.7 Fixed threads parameter to gem-mapper
1.7 Removed Indels Field on Variants Report.
1.7 Added Overlapping Bases at Mapping Report
1.7 Modified Base Counts Overall, removed Base Counts general and Base Counts Overall
1.7 New Dinucleotide CpGs Report
New table dinucleotide stats
New plots for Informative Reads and CpGs
Methylation levels plots for different types of CpGs
Summary Table
1.7 New Readme file to inform about report test
1.7 New basic statis table for Variants Report
1.7 Removed parameter -r (reference length) parameter for mapping reports command (gemBS bsMap).
1.6 New CpGs Density plot, include box plos, bar plot and fitting curve
1.6 Change name at CpG report:
"Heterozygous" for "Alternative CX"
"De Novo CpGs Methylation Status" for "Non Reference CpGs"
"CpGs with SNP" for "SNPs (CX) at Reference CpGs"
1.6 CpGs Report Simplified to Q>20
1.6 BigWig Default parameters for filtering CpG per a given quality and a total number of supported informative reads
1.5 Initial Release
gemBS:
- Marcos Fernandez-Callejo - marcos.fernandez@cnag.crg.eu
- Simon Heath - simon.heath@gmail.com
gem mapper:
- Santiago Marco-Sola - santiagomsola@gmail.com
bisulfite caller and filtering:
- Simon Heath - simon.heath@gmail.com